README.md

metadata

license: mit
tags:
  - geospatial
  - satellite
  - remote-sensing
  - time-series
  - neural-compression

GeoNDC Sample Data

Sample .gndc files for use with pygndc — a neural network-based compression format for geospatial time-series data.

What is .gndc?

A .gndc file stores 4D satellite imagery (Time × Height × Width × Bands) compressed by neural networks, achieving up to 300:1 compression ratios while supporting:

• Reconstruction at any time point (continuous temporal interpolation)
• Fast random pixel access at any timestamp
• Analytic spatial/temporal gradients

Usage

pip install pygndc

import pygndc

with pygndc.open("S2_sample_2022_to_2025.gndc") as ds:
    print(ds)                                        # metadata
    frame = ds.read(t=0)                             # (H, W, C)
    frame = ds.read_at_time("2024-06-15")            # continuous interpolation
    series = ds.sample(lon=116.825, lat=40.486)      # (T, C) time series
    ndvi = ds.ndvi(t=0, red_band=2, nir_band=3)     # (H, W)
    ds.to_tif("output.tif", t=0)                     # export GeoTIFF

Online Viewer

Browse .gndc files directly in the browser (no installation needed):
https://www.geondc.org/viewer/

Links

- pygndc: https://github.com/jianboqi/pygndc
- Website: https://www.geondc.org